• State Resolved
  • Locked Locked
  • Replies 8 replies
  • Subscribers 46 subscribers
  • Views 811 views
  • Users 0 members are here
Support feedback
Options
Options
Similar topics

This thread has been locked.

If you have a related question, please click the "Ask a related question" button in the top right corner. The newly created question will be automatically linked to this question.

ADS 1248 Common Mode Voltage

Jeffry Merkley
Prodigy 70 points

I have a question about the ADS 1248 common mode voltage.  I have attached a document with calculations showing my understanding of the common mode voltage in my application.  I am currently running 300 uA through a 350 ohm resistor and then through a PRT in series and then to ground. I am reading the 350 ohm resistor and the PRT with different analog input channels on the 1248.  The noise level is great at low PRT resistances. Once I get up to around 350 ohms on the PRT, which should be full scale, the noise increases dramatically. My question is whether the analog inputs are being exceeded by the 350 ohm resistor. By my calculations the input on the most positive side of the 350 ohm resistor should be 210 mV at full scale which should not be exceeding the common mode voltage at a gain of 16. I have also attached a schematic of my circuit for reference. The circuit is being run in an AC manner by the analog switch on the left side.

3757.ADS1248 Vcm Calcs.pdf

  • 0
    TI__Guru* 92715 points

    Jeffry,


    It doesn't look like your input are violating the input common mode range. With supplies of +2.5V to -2.5V, and the inputs of 0.210V (+) and 0.105V (-) you should be easily within range for a gain of 16. The output of the amplifiers will be 0.9975V and -0.6825V (also giving the proper 1.68V measurement).

    If your measurments are noisy, but basically correct, I'd look to some other part of the circuit. Often when measurements get noisier with a larger input, it's because the reference noise is large. In your schematic, it's not clear what you are using as the reference.

    If you can, give me a brief rundown on what you're measuring. It would seem to me that this would normally be a ratiometric measurment, but I'm not sure how you have the currents routed to get the proper reference measurement.


    Joseph Wu

  • 0 in reply to Joseph Wu
    Prodigy 70 points

    Joseph,

    Thanks for the quick reply. We are using the internal voltage reference. Sorry I forgot to include that in the original post. We are reading a PRT across channel AIN0/1 and a 350 ohm resistor across AIN2/3.  The current is reversing after a reading from each of these channels. Thanks,

    Jeff

  • 0 in reply to Jeffry Merkley
    Prodigy 70 points

    Joseph,

    Can you also explain a little more about why the noise increases with increasing signal? The way we are getting a ratiometric reading is by taking the reading across the PRT then across the 350 ohm resistor and using a micro to do the ratio. Thanks,

    Jeff

  • 0 in reply to Jeffry Merkley
    TI__Guru* 92715 points

    Jeff,


    Can you describe how you're making this measurement? In the first post you say that the circuit is being run in an AC manner and I don't know what that means. Also in a later post, what do you mean current is reversing after reading.

    Also can you describe the noise that you are getting? Are you getting noise? or is it simply an error? If it is noise, can you explain how it's changing as you change input values? It might help to put up some readings from the ADC.



    Joseph Wu

  • 0 in reply to Jeffry Merkley
    TI__Guru* 92715 points

    Jeff,


    The noise increases with the input signal when the reference is noisy. If you think about what an ADC does, it compares the input to a reference value and basically reports a code that is a percentage of the input value.

    Lets take the two extremes of the input. When the input is 0V, the output code is going to be 0. Even if the input is noisy, it's still going to be 0% of the reference. Now take the input to a value close to full scale. In the is case the measurement will be equal to 100% of the reference. If the input is noise free, but the reference is noisy, you're still comparing the input to a noisy reference so the measurement will have the same noise as the reference. If you have 5 mVpp of noise on the reference, you'll have at 5 mVpp of noise in the measurement when the input is at full scale.



    Joseph Wu

  • 0 in reply to Joseph Wu
    Prodigy 70 points

    Joseph,

    Running the circuit in an AC manner is the same as the current reversal.  The op-amp circuit controls the current supply. The op-amp is given a +2.5 signal to generate the current.  The A/D makes a conversion on the PRT channel and then a conversion on the 350 ohm channel.  The op-amp is then given a -2.5 signal and generates a current in reverse of the first direction.  A conversion is made on the PRT channel then on the 350 ohm channel.  The micro controller then takes these 4 conversions and calculates a resistance based on the ratio between the PRT conversions and the 350 ohm conversions. 

    I am fairly confident that the noise is noise and not an error.  I have attached two graphs of the micro-controllers calculated temperature as well as an excel spreadsheet with the conversions codes from the Tektroniks oscilloscope I am using.

    I understand the two extremes of the noise explanation you gave. Does this apply midway in that the midway voltage would be 2.5mV of noise or would the midway voltage have 5mV of noise as well? I am seeing an incremental increase in noise as the input signal increases.

    Thanks

    Jeff

    0876.120 and 350.xlsx7317.1248@120.pdf5025.1248@350.pdf

  • 0 in reply to Jeffry Merkley
    TI__Guru* 92715 points

    Jeff,

    I did look through the numbers a bit and it does look like noise. I can't tell if there's anything periodic that might be corrupting it (like 60Hz noise) and that might be possible depending on the data rate that you use and your setup (what data rate are you using anyway?).

    For the contribution of reference noise, it typically doesn't rise linearly with the input signal. Due to the nature of the internal workings of the ADC, it starts smaller, and rises faster with larger inputs (I'd describe it as more parabolic). Looking at the standard deviation of each of the four readings collected, it's hard to get a feel for

    I would try to determine if there's some sort of settling error associated with changing from channel to channel. Instead of doing a single read for each of the four settings, can you do 10 or more for each read? It would be good to compare the noise of multiple reads against the noise of each single read collected together.

    Joseph Wu

  • 0 in reply to Joseph Wu
    Prodigy 70 points

    Joseph,

    We are running at 80 SPS. For our current project I have obtained acceptable noise levels. I will probably be doing more experimentation with some of the other ideas you provided for later, more noise sensitive, projects.  I appreciate all your help!

    Jeff